Probabilistic estimates of extreme events are much relevant for the planning of human activities. Thus, probability distributions enable the estimation of maximum annual daily precipitation events, associated with their respective frequencies. The aim of this study was to identify, among the Gumbel distribution, Gamma at 2 parameters and the Generalized Extreme Value (GEV), which is the most suitable to be used on historical series of annual maximum daily rainfall in the Rio Verde basin in Southern Minas Gerais. The parameters of each distribution were estimated by the methods of moments, maximum likelihood and L-moments. For this, eight rainfall stations were used with historical series of at least 20 years distributed over the watershed. By statistical tests as the Kolmogorov-Smirnov, Chi-square, Anderson Darling and Filiben test, the Gumbel distribution obtained the least appropriate adjustments. The GEV distribution by L-moments method was the most appropriate for probabilistic analysis of annual daily extreme precipitation events in the Rio Verde basin and should be prioritized for planning in the region.
extreme events; parameter estimation; statistical inference; statistical tests
